Explainability and Transferability of Machine Learning Models for Predicting the Glass Transition Temperature of Polymers

TL;DR

This paper investigates how machine learning models predict the glass transition temperature of polymers, emphasizing the importance of feature representation, training data range, and model simplicity for transferability and understanding.

Contribution

It demonstrates that simple linear models perform as well as complex nonlinear ones and reveals the impact of training data range on model transferability and the discovery of new polymer chemistry correlations.

Findings

Linear models are as accurate as nonlinear models for Tg prediction.

Transferability improves with broader property ranges in training data.

ML uncovers new structure-property correlations in polymers.

Abstract

Machine learning offers promising tools to develop surrogate models for polymer structure-property relations. Surrogate models can be built upon existing polymer data and are useful for rapidly predicting the properties of unknown polymers. The accuracy of such ML models appears to depend on the feature space representation of polymers, the range of training data, and learning algorithms. Here, we establish connections between these factors for predicting the glass transition temperature of polymers. Our analysis suggests linear models with a smaller number of fitting parameters are as accurate as nonlinear models with a large number of hidden and unexplainable parameters. Also, the performance of a monomer topology-based ML model is found to be qualitatively identical to that of a physicochemical descriptor-based ML model. We find that the transferability of ML models enhances as the property range of the training data increases. Moreover, we establish new Tg polymer chemistry correlations via ML. Our work illustrates how ML can advance the fundamental understanding of polymer structure-property correlations.